The proposed study will follow-up respondents interviewed in the Urban Men's Health Study (UMHS: interviewed Nov. 1996-March. 1998; n=2718). The UMHS has gathered a community based probably sample of men who sex with men (MSMs) residing in New York, Los Angeles, Chicago, and San Francisco. The proposed study addresses the following issues: (1) It is important to identify for intervention MSM population segments who are not seeking treatment, monitoring their health, or complying with treatment regimens. The proposed work will monitor changing trends in health care practices of HIV positive and negative men, and examine barriers to anti-retroviral use. (2) Despite significant risk behavior change over the past decade; many men have been still practice risky sex (see Progress Report). Today, risk behavior occurs in a broader context of treatment programs which are difficult to manage and many MSMs may be motivated to discontinue or inconsistently use anti-retroviral therapy, which, in turn, may increase development of drug resistant HIV strains. It is important then to establish the direction and magnitude of change associated with risk practices and identify which groups show higher rates of increasing risk. Although the UMHS provides a baseline for this monitoring process, a longitudinal study would determine the direction that risk behaviors rates are taking among high risk MSG segments. (3) Risk factors and their distribution are not static in the population. Changing patterns of sexual mixing can substantially modify which segments of the population will be at high risk in the future. Proposed research on modeling sexual mixing effects on HIV transmission, and on identifying bridge populations will expand our ability to predict the flow of infection from high to lower risk groups, and better target for intervention key "mixing" groups. Further, proposed mapping of risk factors and HIV infection would provide a physical guide to targeting prevention efforts. (4) To examine risk behavior among-in and out- migrants, the proposed study will obtain an additional cross sectional sample (n=600). The new sample will also add power and precision to many analyses, and test for secular change and attrition effects.